invalid syntax using cdfplot

[Stephen Okiya]

Hi,

I am using a user written command cdfplot in Stata 15 to generate normalized cdf by a categorical variable using the code

cdfplot x [pweight=weight], normal nocdf by(y)

However, I am getting invalid syntax error. The code works with some categorical variables but results to an error while using other variables.

Kind request for some help on the issue.

Thanks in advance.

Best,
Stephen.

[Nick Cox]

Syntax looks correct. I don't see how we're expected to say more as this is just "works sometimes and not others". We have the same problem with our cooker and don't expect that information to be much help to the repair person.

Give a data example for when it doesn't work, please.

BTW, I wouldn't describe the option normal as producing normalized cdf. The data aren't normalized in any sense; they're shown as they are. The option just superimposes a corresponding normal cdf for the same mean and SD. I would always prefer the results of qnorm myself.

[Stephen Okiya]

Thanks Nick for the reply. Much appreciated.

Below is example of the data

* Example generated by -dataex-. To install: ssc install dataex
clear
input float(weight x) long y
 56.48856 .z  1
   430.42 .z .z
 52.32442 .z  0
2400.9814 .z  1
 42.92722 .z .z
4.0646443 .z .z
 111.2805 .z  0
412.74255 48 .z
2.1733863 .z .z
191.64484 .z  0
 847.1944 .z .z
 4.325087 .z .z
  23.4709  8 .z
31.221676 72 .z
.20838735 50 .z
 49.40934 .z .z
 817.3868 .z .z
 27.78372 .z .z
 856.5333 .z .z
 508.9236 .z .z
124.54675 .z .z
 43.72495 .z  0
  1567.02 .z  1
 171.7089 39 .z
 55.47367 30  1
 85.28024 .z .z
22.181936 .z .z
 60.94157 .z  0
149.82014 .z .z
 2.512651 .z .z
14.018277 .z .z
 42.92722 .z .z
.55545515 .z .z
 76.89115 56  0
 24.82574 21  0
15.335054 .z  0
 97.69048 .z .z
 357.1795 .z  1
 704.0506 .z .z
   2569.6 .z  1
 817.3868  2  0
 6.501606 .z .z
166.73073 74 .z
15.303948 .z .z
 634.8437 .z .z
 1814.716 .z .z
 .6323494 .z  1
 .3003667 .z .z
 46.29515 .z .z
 48.55399 .z .z
end
label values y births
label def births 0 "No", modify
label def births 1 "Yes", modify
label def births .z "Not administered", modify

[Nick Cox]

That's now a reproducible example.

The problem is the missing values on y, which tickle a bug.

On a quick guess, the program ignores the missing values in one place but not in another and ends strangled by an inconsistency. If you don't care about the missing values, you should exclude them explicitly, If you do care about them, you could recode .z to 2, say. Or fix the program. Or write to the author, who is not usually active on Statalist.

[Stephen Okiya]

Thanks Nick for identifying the issue. Indeed when I exclude the missing values the command works.
I really appreciate the constant assistance.

[Stephen Weinberg]

That's now a reproducible example.

The problem is the missing values on y, which tickle a bug.

On a quick guess, the program ignores the missing values in one place but not in another and ends strangled by an inconsistency. If you don't care about the missing values, you should exclude them explicitly, If you do care about them, you could recode .z to 2, say. Or fix the program. Or write to the author, who is not usually active on Statalist.

I was just banging my head against this exact same bug, and this thread saved my %$$%. Thank you, Nick!